Money Management...Any Ideas?

[BJennet]

Originally Posted by Ted Craven

Correct me if I'm wrong, but I always understood that what is debunked is the notion of a progression method (e.g. 'martingale') turning a flat bet loss series into a positive return. Schwartz' HMI (including 'Opponent Method') makes no claim to do so, and even acknowledges that truth. His methods seek to improve on any flat-bet profit and smooth out the wild fluctuations which come with full Kelly proportion betting (and outperform various fractional Kellys).

The HMI spreadsheet template accepts a series of bets (which presumably show a flat bet profit) and proposes fluctuating bet sizes (i.e. modified following Wins and Losses) and compares the resulting bankroll growth to flat-bet, full and fractional Kelly. I confess I have not made use of it in the past 2 years, but my recollection is that HMI outperforms flat betting, and even Kelly varieties.

It sure helps to have a flat-bet profitable method (otherwise I presume HMI merely mitigates your losses), and also a fairly high hit rate (e.g. >= 50%)

I will plug in my last 50 bets and see what it says.

Ted

Hi Ted,

Any system that makes wagers dependent on previous outcomes, and especially immediately previous outcomes - that treats them as anything other than independent events, i.e. Bernouilli trials, is in effect, a type of progression, and thus false. What's especially harmful about them is that they play into gambler's deluded belief in streaks, e.g. the 'Hot Hand' fallacy - that one should win the next race because one has won the previous race, and the 'Gambler's Fallacy' - that one should lose the next race because one has won the previous race.

Therefore, I know Dave's betting system can't outperform Kelly. And BTW, if it could, there are many people on Wall Street and in the gambling world, who have already made a great deal of money using Kelly, and spend much of their time thinking about issues like this, who would love to have something that could outperform it.

Re the notion of 'testing' a betting system, I think it's worth a mention that, unlike most areas of handicapping, money management is strictly based on mathematics, the truths of which are frequently counterintuitive, so it may be advisable for those of us without higher level math skills to be cautious about the conclusions we draw. In the case of HMI, I'm sure you know that 1000 races would be closer to what would really constitute a valid sample for testing. Also, for any betting system, the hit-rate should be irrelevant - only your edge matters.

About the record-keeping part of the software, it all makes perfect sense to me, and I wouldn't disagree with anything Dick said. About hating the races, I take his point - most pro blackjack players acknowledge the tedium of the game, but they're in it for the freedom, the travel, and the need to work only about 500 hours per annum. But I get the sense that most poker players, sports bettors, and horseplayers really love what they do.

Cheers,

B Jennet